Cyberattack on Mexico's Gov't Agencies Highlight AI Threat

**Executive summary:** A small hacktivist group leveraged Anthropic's Claude and OpenAI's ChatGPT—after jailbreaking their guardrails—to automate discovery, exploitation, and lateral movement across at least nine Mexican government agencies, exfiltrating over 195 million identities and tax records plus vehicle and property records; researchers recovered full attacker–AI transcripts and found persistent backdoors and long-term access.